In two previous papers, algorithms designed for searching, matching and identifying phases in qualitative X-ray powder analysis were reported [Schreiner, Surdukowski & Jenkins (1982). J. Appl. Cryst. 15,[513][514][515][516][517][518][519][520][521][522][523][524][525][526][527][528][529][530]. This paper extends the searching technique to unknowns containing isotypical phases and solid solutions. Methods are described for distinguishing among isotypical compounds in qualitative analyses, and for permitting the retrieval of phases in a data base that differ from measured patterns by a uniform Ad/d scale factor.

Several new techniques have been developed for multiphase identification using X-ray powder diffractometry. Among these are (1) automatic correction for common systematic errors, (2) probability-based scoring, (3) window-less searching and matching, and (4) additive phase isolation. These new approaches to old problems are appropriate for implementation on a minicomputer and a program has been written which incorporates them. Results obtained from this work have demonstrated success in solving identification problems under a wide variety of circumstances, many of which were previously considered very difficult for computer algorithms. This paper is the first of two intended to describe the new algorithms and their implementation in the program, SANDMAN.

During the past three years we have undertaken the development of a complete X-Ray Powder Diffraction, facility with the goal of fully integrating experimental and analytical procedures. Such an approach potentially offers substantially improved performance over previously existing systems by virtue of its internal self-consistency and it opens the possibility of significantly extending analytic procedures for both qualitative and quantitative analyses. Our work to date has resulted in improved performance and significant extensions in both areas, and today I will report on those advances in the area of qualitative analysis.

A multiphase identification search/match algorithm based on a probability scoring procedure has been designed and implemented as the kernel of the SANDMAN program. The probability distributions associated with the inexactitudes of the line intensities and d spacings (line positions) are discussed. Methods are developed to narrow the widths of the distributions by separating systematic and random errors. The probability-based scoring algorithm is applied to diffractograms exhibiting substantial specimen displacement error, small broad peaks and large lineintensity variations.

No abstract